Interesting work and I’m really interested to see where this direction goes.
One question I keep coming back to is where tools like this MCP go beyond templating, like what I already use heavily in Ableton Live. In music production, many tasks are repetitive but not identical and that’s exactly where something like this can shine.
At the same time, music is widely seen as a "manual" craft. Every step in the chain from playing an instrument to a final music piece has both a technical and a creative side, and part of the process is staying curious and critical about what could be improved / done differently in the next project. That makes it an open question where automation actually adds value versus where it takes something away.
Where I’d personally love to see AI make a difference is in audio engineering / post processing, which also requires a lot of creativity beside a solid fundament of experience to really excel. There’s often a big gap between a great musical idea and a polished mix or master. If AI could help close that gap and contextually help to improve on things like tone, space, EQ, and loudness would be hugely valuable.
But the key for me is trust and transparency about modifications. I don’t want a black box that just makes things "better™." I’d want something that clearly explains its actions, like: "I've added an EQ to the piano on track 3 at 01:23 to open it up for the bridge so it sits better in the mix."
That kind of assistive, explainable approach would feel much more aligned with how people actually would be open for an assistant to create music.